Drone Scene Reconstruction via Real-Time 3D Perception

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current methods for collecting scene reconstruction data using drones are inefficient due to reliance on pre-calculated paths and geometric prior information, leading to longer reconstruction times and potential safety issues when inaccuracies occur.

Innovation Solution

A method that performs real-time environment perception to detect objects and reconstruct three-dimensional data, determines reconstruction targets based on distance thresholds, and generates exploration paths using path planning algorithms for full coverage, eliminating the need for pre-calculated geometric information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If pre-calculated paths and geometric prior information are used for data collection, then the path planning can be performed offline, but the reconstruction period is greatly increased and safety problems occur when pre-calculated information is inaccurate

Engineering Contradiction:

Improvesafety of data collectionVSAvoidreconstruction period

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent transitions from static offline path planning to dynamic real-time path planning. The drone performs real-time environment perception and reconstructs three-dimensional representation data during flight, allowing the path to be dynamically adjusted based on actual scene conditions rather than relying on pre-calculated geometric prior information

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent performs preliminary environment perception and three-dimensional reconstruction before final path planning. By detecting objects and reconstructing their three-dimensional representation data in advance during the exploration phase, the system prepares necessary scene information to enable efficient subsequent data collection without requiring extensive pre-calculated geometric information

Inventive Principle:

Principle #10Preliminary action

2Productivity

If offline path planning is used with pre-calculated scene information, then the path can be determined before flight, but the collection time is longer and accuracy is reduced when pre-calculated information is inaccurate

Engineering Contradiction:

Improvedata collection efficiencyVSAvoidpath planning accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent implements real-time feedback mechanisms where the drone continuously performs environment perception, detects objects, and reconstructs three-dimensional representation data during flight. This real-time feedback loop allows the path planning to be adjusted based on actual scene conditions, improving both accuracy and efficiency compared to offline planning with potentially inaccurate pre-calculated information

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system performs self-perception and self-reconstruction of the environment. The drone autonomously detects objects and reconstructs their three-dimensional representation data without requiring external pre-calculated geometric prior information, enabling it to serve itself in navigating and planning its own path based on real-time scene understanding

Inventive Principle:

Principle #25Self-service

3Loss of information

If additional flight is performed to obtain geometric prior information, then the scene geometric information can be obtained, but the reconstruction period is greatly increased

Engineering Contradiction:

Improvegeometric prior information completenessVSAvoidreconstruction period

Core Design Contradiction:

Loss of informationVSLoss of time

Solution Approach 1:

The patent merges the path planning process with the environment perception and three-dimensional reconstruction processes. Instead of performing separate additional flights to obtain geometric prior information, the system combines these functions into a single integrated workflow where the drone performs real-time perception, reconstruction, and path planning simultaneously during its data collection flight

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent maintains continuous useful action by performing environment perception, object detection, and three-dimensional reconstruction continuously during the flight path exploration. This eliminates the need for additional separate flights to gather geometric information, as the reconstruction process occurs continuously alongside the data collection mission

Inventive Principle:

Principle #20Continuity of useful action

Data Source

PatentUS11640696B2Method, computer device, and storage medium for exploring and collecting scene reconstruction data

Publication Date: 2023.05.02 SHENZHEN UNIV

AI summary

A method, computer device, and storage medium for exploring and collecting scene reconstruction data. The method includes: capturing a scene image; performing real-time environment perception on the scene image to detect object to be reconstructed in the scene image and reconstructing three-dimensional representation data of the object to be reconstructed; determining a reconstruction target in the object to be reconstructed based on the three-dimensional representation data of the object to be reconstructed and a preset distance threshold; generating an exploration path according to a preset exploration path planning algorithm and exploring an area to be reconstructed according to the exploration path; and collecting reconstruction data of the reconstruction target according to the preset exploration path planning algorithm when the reconstruction target is reached.